What Assessment Can and Cannot Do

This is a summary of a paper by Dylan Wiliam originally published in Swedish in Pedagogiska Magasinet in September 2011 which can be downloaded from http://www.dylanwiliam.net/


There has to be a system for holding educators accountable to those who pay. High-stakes assessment can be used for this accountability. Strong evidence suggests that "the presence of a high-stakes accountability system raises student achievement by the equivalent of as much as an extra two months’ learning each year" but every high-stakes assessment system studied has negative unforeseen consequences. This is because (a) less than one quarter of student achievement can be attributable to the school and (b) 'Campbell's Law': the use of a quantitative social indicator distorts the social phenomenon it is intended to measure eg teachers teaching to the test ensure that pupils get better at passing the test but often at the expense of them getting worse at things that are not tested. 


Furthermore, we tend to measure what is easy to measure rather than what should be measured. Thus in languages we assess written proficiency more than spoken proficiency because it is easier to measure. "Even if the teachers try to take a broad view of the curriculum, the students will often resist, asking, 'Will this be on the test?'"


The ideal assessment system would be:

• Externally referenced ie not dependent on the standards of each individual teacher
• Distributed across the whole course
• Cumulative ie what is learned at one point of time will be needed later so that students cannot just cram for the test and then forget everything they have learnt.

Assessment improves learning when:
It helps teachers and learners to clarify, share and understand what is to be learned and how we know it has been learnt
It helps engineer "classroom discussions, activities, and tasks that elicit evidence of student achievement"
It provides formative feedback
It encourages students to collaborate acting "as learning resources for one another"
It encourages students to be "owners of their own learning"


But "weighing the pig doesn't fatten it".


Furthermore, AfL can be taken to extremes. "giving grades and scores too frequently will certainly slow down learning, but not giving students any indication of whether they are making progress is just as misguided. The important thing about feedback, however it is given, is that it should cause thinking. Once an emotional reaction occurs, the learning will certainly stop, but there are ways of telling students whether they are, indeed making progress that does not allow them to compare their current achievement with others, thus minimizing the extent to which the student reacts to the feedback by attending to their well-being rather than by using the feedback to improve."